Motion detectors typically employ passive infrared (IR) sensors that detect the heat from moving persons or other warm objects and generate an electrical signal. Such detectors typically include a pyroelectric material and a multi-modulating optics, often referred as Fresnel lens, alternatingly focusing light on two pyroelectric elements. The pyroelectric elements generate an electrical signal if the incoming heat flux changes over time. The pyroelectric detector thus acts as natural electrical high-pass by being sensitive to motions occurring above a certain typical frequency range. Depending on the sensing element size, the cut-off frequency may be as low as 0.4 Hz for large element sizes or higher for smaller elements. Typically, motions of a person occur in the range of approximately 0.4 Hz to 4 Hz, so elements and signal processing electronics in proximity detectors are typically tuned to this range. For decades, pyroelectric infrared (IR) detectors were considered the easiest approach to motion sensing for light switches and alarm units since they deliver high signal levels which could be processed by then available analog electronics.
More recently, a single pixel thermal sensor has been developed to detect a frequency response down to steady-state heat flux (direct current (DC)) in conjunction with a multi-element modulating optics which modulates the signal over the total field-of-view (FOV) of the sensor. Such a device is able to detect motions within a much broader frequency range, even down to DC, which corresponds to the detection of stationary objects (“Presence Detection”) or to higher frequencies, which additionally allows the detection of faster human bodily motions, such as hand waving or other gestures
Previously, gesture recognition techniques have generally been divided into imaging and non-imaging techniques. The imaging techniques typically employ multi-pixel sensors with optics that map the FOV onto the different pixels, so motions and gestures can be evaluated by means of image processing methods. While many of the imaging methods evaluate motion in a two dimensional plane, imaging may also be performed in three dimensions by including depth information, which can be achieved by a variety of methods such a time-of-flight, stereo images, structured light pattern recognition, or others.
Fewer non-imaging gesture recognition techniques have been employed. One non-imaging technique utilizes an electric field, in which change is detected by means of a capacitive detector array in the sensing plane. Another non-imaging method employs the reflection of an infrared beam sent out by an IR light-emitting diode (IR LED). The beam is reflected off an object and is detected by one or more photodiodes. These non-imaging methods incorporate multi-pixel or multi-electrode sensors. The non-imaging solutions do not make an image of the scene, since there is not a defined relationship between a specific FOV segment with a certain detector pixel.
Analysis of frequency patterns in the output of non-imaging sensing devices is known. For example, inertial sensors, as commonly employed in mobile device for the detection of motion of the device, may have software that looks for frequency and amplitude patterns and for the fingerprint of a certain detected physical motion. As such, the software can determine, for example, if the user carrying the mobile device is walking, driving in a car or on a train. It is also possible to determine if the device is lifted up and placed at the ear of a user to take a call, solely by comparing frequency and amplitude pattern of the inertial sensor output signal with those in a library of gesture signatures. Such pattern recognition software may be self-learning, and the library can be extended or adjusted by extracting common patterns from other behavior of a user. However, the output of non-imaging sensors has been insufficient to recognize multi-dimensional movement or gestures.
Therefore, there is a need in the industry for a movement or gesture recognition solution that addresses at least some of the abovementioned shortcomings.